物件導向系統之有效率的動態記憶體配置器

Efficient Dynamic Memory Allocators for Object-Oriented Systems

姚立展(Li-Jhan Yao)

指導教授 ： 鄭福炯

大同大學/電機資訊學院/資訊工程學系所/碩士(2006年)

摘要

過去有許多研究者集中於利用硬體實作的方式，來改進傳統夥伴系統 (binary buddy system) 的效能，但都無法非常有效地增進其效能。本篇論文提出三個可配置任何記憶體空間的硬體協助記憶體配置器，使得夥伴系統配置記憶體的空間盲點，可以被消除。這三個可配置任何記憶體空間的硬體協助記憶體配置器，有效的消除了記憶體的外部碎裂問題。我們利用SPECjvm98 的標準檢查程式來做模擬。模擬的結果顯示，可配置任何記憶體空間的硬體協助記憶體配置器，使用了較少的記憶體空間，並有效的增進了記憶體的利用度。

關鍵字

記憶體 ； 配置器 ； 利用度

並列摘要

Many researchers have been focused on improving the performance of the buddy system via a hardware implementation, but that does not improve performance drastically. This thesis presents three all size sets hardware-assisted memory allocators (HAMAs) based on Chang’s or-gate tree to eliminate the size blind spot in the buddy system. Our HAMAs can allocate a free memory block of any size in the system. This effectively eliminates external fragmentation. We use SPECjvm98 benchmark to collect the information of object allocation and de-allocation and then exploit this information to test our HAMAs and others. The results of simulation indicate that our all size HAMA uses less memory space and effectively improve memory utilization.

並列關鍵字

Java ； SPECjvm98 ； buddy system ； allocator ； memory

參考文獻

[1] Donald E. Kunth, “The Art of Computer Programming,” Addison-Wesley, Jan. 1973

[2] Kenneth C. Knowlton, “A Fast Storage Allocator,” Comm. ACM, vol. 8, pp. 623-625, Oct. 1965

[3] James L. Peterson and Theodore A. Norman, “Buddy Systems,” Comm. ACM, vol. 20, pp. 421-431, June. 1977

[4] Kenneth K. Shen and James L. Peterson, “A Weighted Buddy Method for Dynamic Storage Allocation,” Comm. ACM, vol. 17, pp. 558-562, Oct. 1974

[5] Daniel S. Hirschberg, “A Class of Dynamic Memory Allocation Algorithms,” Comm. ACM, vol. 16, pp. 615-618, Oct. 1973